Radiance Assimilation over
Northern High Latitude Regions
Zhiquan Liu, Hui-Chuan Lin, Thomas Auligne
(NCAR/MMM)
Acknowledgements:
Collaborators from Ohio State University
4/19/2010
ITSC-17, Monterey
1
Outline
• Background: Arctic System Reanalysis
(ASR) Project
• Preliminary Results from 2 test periods
• Summary and future work
4/19/2010
ITSC-17, Monterey
2
Background
• Arctic System Reanalysis (ASR) project
–
–
–
–
–
Funded by US NSF
University efforts: OSU, NCAR, UIUC, CU
11 years reanalysis: 2000~2010
Currently testing for 30km, may go up to 10km?
NCEP provided conventional and radiance data.
• Testing/Tuning the system for 2 months
– Dec. 2007: use NCEP FNL (1°X1°, P levels) as LBC
– Aug. 2008: use ERA-Interim (80X80km, model level) as LBC
4/19/2010
ITSC-17, Monterey
3
WRFDA-3DVAR
•
WRFDA includes 3D/4DVAR and Hybrid VAR/Ensemble
scheme
•
3DVAR is adopted for ASR for computational efficiency
– 3-hr cycling regional DA, time window: ±1.5h
•
Control variables: stream function, unbalanced velocity
potential, unbalanced T, pseudo RH, unbalanced Ps.
•
NMC method to generate background error covariance
statistics.
– Domain-averaged statistics.
•
Recursive filter in horizontal, EOF in vertical for covariance
modeling.
4/19/2010
ITSC-17, Monterey
4
ASR domain
WRF model version 3.1:
30km*30km (360*360),
70 Levels (Top@10hPa)
40m-50m vertical spacing in PBL
Sea-Ice
4/19/2010
DFI, GWD, fractional sea-ice
WSM5 MP, new Grell,
MYNN2.5 PBL, sw/lw RRTMG,
Noah LSM.
ITSC-17, Monterey
5
NCEP PREPBUFR data (2007120100, +/-3h time window)
4/19/2010
ITSC-17, Monterey
6
GPSRO data caused
large cold biases near
model top due to bad
quality control (data
above 30km not rejected).
Will re-include in future
exps. with refined QC.
4/19/2010
ITSC-17, Monterey
7
Radiance data used
amsua
amsub mhs
Noaa-15
4,5,6,7,8,9
3,5
Noaa-16
4,5,6,7,8
3,4,5
Noaa-17
3,4,5
Noaa-18
4,5,6,7,8
3,4,5
Metop-2
4,5,6,8,9
3,4,5
Aqua
4,5,6,8,9
Follow NCEP provided instrument/Channel availability table
AMSU-A ch4 only over sea
Thinning to 90km.
CRTM, Variational Bias Correction
AIRS new cloud detection scheme was found instable.
May include it in future runs.
4/19/2010
ITSC-17, Monterey
8
Results from Dec. 2007
NCEP FNL as LBC
4/19/2010
ITSC-17, Monterey
9
Radiances better positioning Low Pressure centers
W/O Radiances
With Radiances
CASE: LIU
CASE: CYC7
ASR(blue) v. FNL(red): P’@model level-1 @day-20 in Dec. 2007
4/19/2010
ITSC-17, Monterey
10
Precipitation (Monthly total in Dec 2007)
(Unit: mm)
ERA-Interim
4/19/2010
ASR with Radiances
ITSC-17, Monterey
11
Precipitation (Monthly total in Dec 2007)
(Unit: mm)
ASR with Radiances
4/19/2010
ASR w/o Radiances
ITSC-17, Monterey
12
Results from Aug. 2008
ERA-Interim as LBC
4/19/2010
ITSC-17, Monterey
13
ERA-Interim Fractional Sea-Ice
Beginning of August
4/19/2010
End of August
ITSC-17, Monterey
14
500hPa Height Analyses
4/19/2010
ITSC-17, Monterey
15
Monthly Mean T2m
W/O Radiances
With Radiances
4K colder than ERA
4/19/2010
ITSC-17, Monterey
16
Analyses vs. SYNOP (RMSE)
Rad. Improve
T2 fit
Over fitting
To obs?
Rad. degrade
Ps fit
4/19/2010
ITSC-17, Monterey
17
Analyses vs. SYNOP (Biases)
Rad. reduce
Q2 biases.
4/19/2010
ITSC-17, Monterey
18
Z: ASR vs. ERA-Interim
W/O Radiances
4/19/2010
With Radiances
ITSC-17, Monterey
19
Summary
•
Several issues in WRFDA were found during testing
– GPSRO QC, AIRS cloud detection scheme.
– Errors near the model top increasing with the time.
– Other issues not listed/mentioned in this talk ….
•
Radiance impact is mixed
–
–
–
–
–
•
Adding radiances tends to produce more precip. over ocean.
Better positioning Low pressure centers over ocean
Radiances apparently improve T2m/Q2m.
But larger Ps error.
Mixed impact for upper air. Model top biases make things
complicate.
Difficulties to do reanalysis for university.
– Computational resource limitation, limited man power
– Data collection/processing
– Lack detailed obs monitoring information ( usually available in
operational NWP centers. e.g., blacklist of various obs types)
4/19/2010
ITSC-17, Monterey
20
Future work
• More testing/tuning needed before production
run
– Diagnose HBH^T and compare to R
– Use observation sensitivity tools to identify impact
from different obs types (& individual radiance
channel)
– Model top issue: nudging global fields at top
levels? increase model top? Add O3?
• Need run forecasts to judge analysis
performance.
4/19/2010
ITSC-17, Monterey
21
T: ASR vs. ERA-Interim
W/O Radiances
4/19/2010
With Radiances
ITSC-17, Monterey
22